The contrast between permissions to act and permissions to believe. Javier González de Prado Salas a. Universidad Autónoma de Madrid, Spain

Transcription

1 The contrast between permissions to act and permissions to believe Javier González de Prado Salas a Universidad Autónoma de Madrid, Spain *This is a penultimate draft of an article to appear in Philosophical Explorations. Please refer to the published version. Abstract: There is an interesting contrast between permissions to act and permissions to believe. Plausibly, if it is permissible to believe something from a perspective with incomplete evidence, it cannot become impermissible to believe it from a second perspective with complete evidence. In contrast, it seems that something permissible to do for an agent in a perspective with limited evidence can become impermissible in a second perspective in which all the relevant evidence is available. What is more, an agent with incomplete evidence may be permitted to do something that she knows would be impermissible if she occupied a perspective of complete evidence. In this paper I argue that this contrast is explained by a disanalogy between the role played by belief in epistemic deliberation and the role played by action in practical deliberation. Epistemic deliberation may be closed by adopting other attitudes than belief (e.g. accepting some hypothesis as the most likely one), whereas in general practical deliberation can only be closed by endorsing some course of action. Thus, when there are pressures to close some practical deliberation, agents have to make a decision about what to do even if they lack relevant information. By contrast, lacking relevant evidence, agents may always refrain from forming a belief and close instead their epistemic deliberation by adopting some other attitude. Keywords: practical permissions; doxastic permissions; perspectivism; objectivism; practical deliberation; epistemic deliberation. a jgonzalezdeprado@gmail.com

2 1. Introduction There are many situations in which agents with partial information deliberate about what is permissible to do or believe. Does the answer to such deliberative questions depend on the epistemic perspective of the agent (on what evidence is available to her)? Should reasons beyond the agent s epistemic ken have a bearing on what beliefs and courses of action are permissible for her? It is very plausible, I think, that the answer to these questions is different in the case of permissions to believe and in the case of permissions to act. Permissions to act and permissions to believe are affected in different ways by reasons beyond the agent s epistemic ken. In this paper I first present this contrast (and motivate its plausibility) and, second, I offer an explanation for it by appealing to an interesting disanalogy between epistemic and practical deliberation. 1 I will focus specifically on deliberative questions about what is permissible to do or believe that is, on deliberative questions about what actions it is permissible for the agent to perform and what beliefs it is permissible for the agent to hold. I will assume that it is permissible to do (or believe) those things that you may do (or believe), while it is obligatory to do (or believe) what you ought to do (or believe). I will think of oughts and mays in terms of normative reasons which I will take to be facts that count in favour of performing an action or holding an attitude (see Scanlon 1998; Parfit 2011; Álvarez 2010). In this way, I will assume that it is permissible to do (or believe) something if and only if the balance of all relevant reasons sufficiently supports doing it (or believing it). 2 I say relevant reasons because, as I will discuss presently, it could be that some permissions are determined not by all existing reasons but only by those within the agent s epistemic ken.

3 Imagine the following situation. Peter has a very annoying illness. Peter s doctor, Mary, knows a partial remedy for it, but she does not know about the existence of a flower in a nearby mountain that would cure Peter s illness completely (the healing powers of the flower would disappear if the partial remedy is applied). Given that Mary is (blamelessly) unaware of the fact that the mountain flower would cure Peter s illness, should this fact be relevant for the determination of Mary s practical permissions? One possible view is that it should not, because only evidence within the agent s epistemic ken determines what the agent is permitted to do. Call this sort of view perspectivism. 3 An alternative position is that all existing reasons, including those beyond the agent s epistemic ken, can be relevant for determining the agent s practical permissions. Call this view objectivism. 4 Analogous positions can be specified for permissions to believe. In this paper I defend the plausibility of a version of perspectivism about practical permissions, while endorsing a form of objectivism about permissions to believe. More precisely, my claim will be that permissible beliefs have to be sufficiently supported by the balance of all existing reasons (including those beyond the agent s epistemic ken), whereas it can happen that a nevertheless permissible action would not be sufficiently supported if reasons beyond the agent s epistemic ken were to be considered. Put in different terms, the idea is that if it is permissible to believe something from a perspective with limited evidence, it will still be permissible to hold such a belief from a perspective in which all relevant evidence is available (all facts are known). In contrast, an action that an agent is permitted to perform when in a certain perspective (i.e. given the evidence available in that perspective) may become impermissible in a perspective in which all facts are known. This may be so even if all the evidence in the first perspective was good (not misleading or defeated evidence).

4 Notice that I am considering a rather restricted type of objectivism about permissions to belief. What this form of objectivism requires is that such permissions are sufficiently supported by the balance of all existing reasons. But this is only a necessary condition on permissions to belief, not a sufficient one, and it is compatible with further requisites for instance, the requisite that permissible beliefs have to be sufficiently supported as well by the balance of those reasons accessible from one s perspective (see Schroeder 2015b). Thus, permissions to believe (in the picture I will present) may have a perspectivist dimension. Yet there is a clear contrast with the sort of perspectivism characteristic (on the view I will explore) of permissions to act, according to which an action can be permissible even if there exist reasons beyond the agent s epistemic ken that would tip the overall balance of reasons against such an action. Far from being an inconsequential, minor difference, this contrast points towards an interesting divergence in the behaviour of practical and doxastic permissions, as I purport to show in this paper. The two views contrasted can be formulated as follows: Doxastic Objectivism (DO): a belief is permissible only if it is sufficiently supported by the balance of all existing reasons. Practical Perspectivism (PP): an action is permissible only if it is sufficiently supported by the balance of the reasons within the agent s epistemic ken (but it need not be sufficiently supported by the balance of all existing reasons). As I will understand the terms, the best thing for an agent to do (or believe) is that which is supported by all existing reasons: something cannot be the best thing for an agent to do, in this sense, if it would be impermissible for her to do it were she to occupy a perspective with complete evidence. Thus, the claim I will defend is that an agent having limited evidence may be permitted to do something which is not the best thing for her to do whereas an agent with

5 limited evidence is never permitted to believe something false, that is something that is not the best thing to believe (i.e. something that would not be permissible to believe if the agent had complete evidence). I will try to defend and explain this contrast by examining a disanalogy in the roles of belief and action in epistemic and practical deliberation. My focus will not be so much on providing conclusive arguments in favour of PP and DO (although I will give some motivation for these views); rather, my goal will be to provide a convincing account of the sources of this contrast. In the absence of an explanation of why practical and doxastic permissions behave differently, one could think that a unified account of permissions is preferable. Providing such an explanation, therefore, will add to the plausibility of the contrasting views considered. At any rate, given that both PP and DO are independently attractive views (as I will try to show) and many authors are committed to them, it will be interesting to look for possible explanations for such contrasting behaviour between practical and doxastic permissions. Two clarifications. First, it could be argued that there are different senses in which an action or belief can be permissible that is, in an objective or a subjective sense (see Sepielli forthcoming; Dorsey 2012). However, even if there were such an ambiguity, it would not dissolve the debate I am interested in: I will focus on deliberative permissions and obligations, that is, the sort of permissions and obligations one tries to find out about when engaging in deliberations the sort of permissions and obligations one looks for as guidance for one s decision-making (see Kiesewetter 2011). My purpose is to discuss whether and why the correct answers to these deliberative questions about what to do depend only on accessible reasons. Second, the sort of perspectivist proposal I will consider is that permissions and obligations are determined by those (real, objective) reasons which are epistemically accessible to the agent and I will take such reasons to be constituted by facts. So, I will not be defending here the view that what is permissible to do may be determined by misleading evidence or other

6 considerations that merely appear to be good reasons. It may be the case that misleading evidence is relevant for fixing practical permissions, but I will not discuss this question here. 5 The view I will consider is perspectivist because it focuses only on those actually good reasons which are accessible from the agent s perspective, not because it regards as relevant any consideration that merely seems to be a good reason from the agent s perspective. 2. Perspectivism about practical permissions I will start by presenting and motivating PP. As I said above, I assume that reasons are facts, and that practical permissions are determined by what reasons there are. However, I will hold that practical permissions depend on the agent s epistemic perspective, on what reasons are accessible to the agent (in some sense of accessible). So, it is not all reason-giving facts that are relevant for determining the practical permissions of an agent, but only those reason-giving facts that are within the agent s epistemic reach (I will not discuss in detail what this epistemic reach amounts to. You may think of it as those facts that the agent is supposed to acquire knowledge about, given the evidence available). 6 In this way, in the picture I will assume, practical permissions are determined by the balance of the reasons within the agent s epistemic ken. The main motivation, I think, for accepting this picture of practical permissions is provided by examples like the following which is a reformulation of the sort of case presented by Jackson (1991; see also Parfit 2011 for the analogous mine shaft example): MEDICINE: Jill is a doctor treating John, who suffers from some non-lethal, but annoying condition. Jill has access to three different drugs. She knows that drug A will partially cure John (although much annoyance will remain). Drug B will cure John completely and drug C will kill him. Jill knows that one of B and C will cure the

7 condition, while the other will kill John, but she does not know which is which (and she has no way of finding out). If Jill fails to prescribe some drug, John s condition will quickly worsen. It is very plausible to think that Jill is permitted (and obliged) to prescribe drug A, even if she knows that that is not best possible option in the sense that it will not cure John completely, whereas either B or C would (Kolodny and MacFarlane 2010; Zimmerman 2006; Lord 2015; Kiesewetter 2011; Jackson 1991). According to this view, Jill s permission (and obligation) to prescribe drug A is determined by those reasons she has access to (she knows that drug A will relieve John), but not by those reasons that remain beyond Jill s epistemic ken (she cannot know that B is the cure). Perhaps it could be argued that prescribing drug A is actually impermissible (because there are non-outweighed reasons against it), even if Jill is blameless in doing so. However, note that if this were so Jill could be blameless despite knowing that she is doing something impermissible. While it is plausible to think that agents can be blameless in doing things they reasonably take to be permissible but are actually impermissible, it is much less plausible to think that agents can be blameless when doing what they know is impermissible especially since I am dealing with a deliberative notion of impermissibility, that is, a notion of impermissibility that is supposed to guide the agent in finding a correct answer to her deliberative question about what to do (although see Dorsey (2012) for discussion). It seems, thus, that the natural thing to say is that in MEDICINE Jill is blameless because she is doing something permissible. So, in this intuitive reading of MEDICINE, the permissible thing to do is different from the best thing to do, even if the agent knows that what she is choosing to do is not the best thing to do (remember that I am defining the best thing to do as that which is supported by the balance

8 of all existing reasons). From a new, enriched perspective in which Jill gets to know that drug B is the cure, prescribing drug A will become something not permitted: what is permissible to do in this new perspective is to prescribe drug B. Thus, an action permissible given the evidence available in a first perspective may turn into something impermissible in a second perspective in which all relevant evidence is available (all relevant reasons are accessible). MEDICINE, thus, seems to offer support for PP: an action does not need to be supported by the balance of all existing reasons in order for it to be permissible for an agent occupying a perspective with limited evidence. Of course, not everyone accepts this reading of MEDICINE (see, for instance, Graham 2010; Bykvist 2011). However, I will assume that this kind of examples manage to show that practical permissions are determined only by those reasons within the agent s epistemic reach. In the next section I will motivate the plausibility of a contrasting view concerning doxastic permissions a view according to which permissions to believe cannot be lost when reaching a perspective of complete evidence. After that, I will try to provide an explanation of why practical and doxastic permissions differ in this respect. 3. Objectivism about doxastic permissions Let us focus on epistemic deliberation, in particular deliberation about what to believe. I will endorse a version of objectivism about what to believe, according to which one is permitted to believe something only if such a belief is supported by all facts within and outside the agent s epistemic ken. In this way, I will take it that it is not permissible to believe something false. Although I will assume this view without arguing for it, let me provide some motivation for it. It is commonly accepted that it is correct to believe something only if it is true. It is also plausible to think that there is a close connection between what is correct to believe and what

9 one is permitted to believe. In particular, it seems plausible to think that one is not permitted to believe incorrect beliefs. 7 If this is granted, then one gets the view that one may believe only true things: it is not permissible to believe falsities (this view is common in the literature, see for instance: Whiting 2010; Williamson 2000; Shah and Velleman 2005; Boghossian 2008; Littlejohn 2012). 8 Thus, if I find out that my past belief was false, I will regard it as incorrect: it was wrong for me to hold that belief, even if perhaps I was rational and epistemically blameless in doing so (say, it seemed I had very good evidence). Likewise, I will assess somebody holding a false belief as believing what she ought not to believe, even if she lacks the evidence showing the falsity of her belief. As I said, I will accept all these points without arguing for them the considerations above should be taken just as providing some motivation for the view I want to present. Arguably, when an agent knowing all facts is not permitted to believe some claim, it is because such a claim is not true (since that agent would know all true claims and, very plausibly, knowing some claim makes it the case that one is permitted to believe it). Assuming a nonrelativist view of truth, 9 if a claim is actually false from a perspective of complete evidence, it will also be actually false from a perspective in which the agent only has partial evidence and thereby the agent will not be permitted to believe that (false) claim from perspectives with partial evidence. Thus, if it is actually permissible for an agent with limited evidence to believe something, it will still be permissible to believe it from an enriched perspective in which the agent knows all relevant facts: access to complete evidence cannot make impermissible to believe something that, from a less well informed perspective, was (actually) permissible to believe. DO, thus, seems to be correct at least if it is granted that it is not permissible to believe false things and that it is permissible to believe something known (and that the relevant beliefs are not relativist).

10 The view that false beliefs are impermissible entails a (weaker) claim that will be important in the remainder of the paper, namely the claim that it is impermissible to endorse a belief that one knows is not true (a belief against which one knows there are sufficient reasons). One cannot be permitted to have beliefs one knows to be false. I think it is rather implausible to argue otherwise, even for those who do not accept that it is always impermissible to believe something false. So, there cannot be cases in which an agent is permitted to believe something despite her knowing that, if she had complete evidence, she would not be permitted to have such a belief. In other words, there are no scenarios analogous to MEDICINE for belief. Imagine that I am facing a very complicated mathematical problem. There is one approximate solution to the problem, which captures most interesting aspects of the exact solution. However, I know that the approximations made introduce certain inaccuracies in this solution (it is not completely right). In this case, I cannot believe that this is the exact solution for the equation. What I can do is to treat this solution as right in most relevant respects. I can believe that the solution is a good approximation, but I cannot believe its truth. In some cases, it may be that my epistemic deliberation needs to reach some positive conclusion other than withholding belief. 10 (Say, I am in charge of some group of experts advising a policymaker; or I am a doctor and have to prescribe some treatment). So, what happens when I need to conclude my epistemic deliberation whether p but I am not in a position to believe either p or not-p as true? Well, in such cases I can still bring my deliberation to a conclusion by adopting some epistemic attitude (with action-guiding potential) other than belief. For instance, I may assign some degree of likelihood to different claims (and perhaps endorse the claim with the highest degree of likelihood as the most probable one); I can make a guess or a conjecture; I may assume some view for the sake of further investigation; or I can take some claim as a good approximation to the truth an approximation with useful practical applications.

11 All these attitudes can be permissible when there are not sufficient reasons for holding some claim as true for believing some claim. Perspectivism seems right about many of these epistemic attitudes: arguably, the permissibility of adopting many attitudes of this sort depends only on the reasons accessible to the agent. So, even if one is permitted to believe only truths (and in particular one is not permitted to believe what one knows to be false), there are other epistemic attitudes that can be permissibly adopted when one is not ready to think of any claim as the truth about the matter under deliberation what is more, even when one is certain that the truth about the relevant matter is unknown to oneself. Thus, no matter how much pressure I have to bring my epistemic deliberation to an end, I may be able to do so even when I am not permitted to hold any belief about the question deliberated. I can regard some claim as the most likely one, accept some view as a good approximation or make an informed guess. 4. Explaining the contrast Assume that PP and DO are right. The question that arises is why there is such a contrast between the behaviour of doxastic and practical permissions. In this section I provide an explanation for such a contrast, by taking into account an interesting disanalogy in the roles played by belief and action in epistemic and practical deliberation respectively. I have just argued that, even in cases in which epistemic deliberations (say, about whether p) must be brought to a conclusion, agents may do so without endorsing any belief about the matter judged (nor the belief that p, neither that no-p). The agent may conclude her epistemic deliberation by adopting some attitude other than belief. Even if some answer is demanded (not mere suspension of judgment), the agent may endorse some claim as the most likely or promising one, and act on the assumption that it is true.

12 The situation is different in practical deliberation. In many cases, this sort of deliberation has to end with the endorsement of some course of action 11 (perhaps leaving things as they stand, which may be seen as a course of action). So, in MEDICINE, Jill is required to do something: she must prescribe one of the drugs. The only (rational) way in which she may conclude her deliberation is by endorsing some course of action. There is no alternative attitude other than endorsing some course of action that will allow Jill to conclude her deliberation appropriately. Of course, practical deliberations do not always have to be brought to a conclusion. On many occasions, the agent will be free to leave the deliberation without endorsing any course of action for instance, she may choose to wait and see how matters develop before making a definitive decision. But, if an instance of practical deliberation is to be concluded, it has to be by endorsing some course of action. Furthermore, in many cases deliberations cannot be deferred indefinitely, but have to be concluded at some point perhaps due to time constraints or other pressures. This is what happens, as I just pointed out, in MEDICINE: a decision must be made, even if not all facts relevant for the decision are known (it is not known whether drug B or C is the cure). Note that the distinction between epistemic and practical deliberation is not just that leaving things unchanged often counts as undertaking a course of action, while not believing anything about whether p (e.g. suspending or withholding belief) does not count as forming a belief. 12 It is easy to think of scenarios both in epistemic and practical deliberation where the agent must do something else than just suspending belief or leaving things as they stand that is, scenarios where the agent must provide some positive answer to the deliberative question. Imagine, for instance an expert that is expected to give some advice to a policy-maker (epistemic case), or the doctor in the MEDICINE scenario above (practical case) remember that in MEDICINE not prescribing any treatment is clearly wrong, since it will lead to a quick

13 worsening of the patient s condition. Even in these scenarios, where the option of leaving things as they stand is ruled out, there is still a crucial difference between epistemic and practical deliberation: when there are pressures to bring the deliberation to an end, in the epistemic case it is possible to do so by adopting some attitude other than belief (e.g. accepting some view as a good approximation or making an informed guess), whereas in the practical case the only way of concluding the deliberation is by endorsing some course of action. Thus, in practical deliberation, agents are often required to choose some course of action even if they lack epistemic access to all the facts relevant to the matter deliberated (even in situations where they are aware that they are not in a position to know facts that would, if known, affect their decision). This is why, I think, the normative standards that regulate rational actions are only concerned with those reasons accessible to the agent. Accordingly, the goal of rational action (assume the good is such a goal) becomes gradable: there are better and worse things to do, and your practical obligation is to do what is best in light of the reasons accessible from your perspective. Perhaps there is some notion of overall practical permission (determined by all facts), but the sort of practical permission one is after in practical deliberation is in the perspectivist view I am arguing for just concerned with those reasons accessible to the agent. If this were not the case, and reasons determining practical permissions could be beyond the agent s epistemic reach, there would be many situations in which the agent is required to conclude a piece of practical deliberation (due to time constraints or other pressures) but she finds herself without rational guidance about how to reach such a conclusion this would happen in cases where the agent knows that relevant evidence is beyond her epistemic reach (e.g. in MEDICINE). 13 In deliberations about what to believe, by contrast, it is not possible to find examples like MEDICINE, in which an agent considers (rationally) that she is permitted to believe something that she knows is false (i.e. something that she know is not the best thing to believe).

14 If the agent knows that the true answer to some question lies beyond her knowledge, she can (and ought to) decline to form any specific belief about the answer to that question. If necessary, she can adopt some other attitude: for instance, if some answer is needed in order to make a practical decision regarding the issue discussed, the agent may make a guess or adopt some attitude with suitable practical consequences (e.g. take some answer as more probable, and choose the course of action that maximizes expected value). The standard governing belief, thus, may remain concerned with all reasons, even those not accessible for the agent. Truth, the goal of rational belief, is not gradable: something is either true or false. One may think that an epistemic version of MEDICINE is the following. Imagine that a coin is tossed, and you do not see how it falls (see Feldman 1988). You are asked about how the coin landed. What is the right attitude for you to adopt? It seems that it is permissible for you to suspend belief about whether the coin landed heads or tails, or perhaps to assign 50% chance to both outcomes (similar examples may be constructed with lottery scenarios). In this case, it seems that it is right to adopt a certain attitude (suspending belief, or assigning equal probabilities) even if one knows that it is not the best attitude that could be adopted that is, the one that would be right to adopt if all facts were known. The best attitude possible, in this sense, would be to believe the truth about the coin toss (say, that it landed tails up). 14 Even if it is granted that this reading of the coin toss example is correct, it does not show that there are cases in which it is permissible to believe something known to be untrue. What it shows, at best, is that there are cases in which an agent is permitted to adopt a certain epistemic attitude other than belief even if she knows that the truth about the relevant matter is unknown to her. I have already acknowledged that there are epistemic attitudes that may be rightly adopted towards propositions not known to be true, even towards propositions known to be false and that, arguably, perspectivism is correct about some such attitudes. But there are no situations in which one may believe what one knows to be untrue, whereas there are

15 cases where it is right to do what one knows is not the best thing that could be possibly done (i.e. what would be right to do if all facts where known). A possible objection is that outright belief and epistemic attitudes such as taking something to be probable do not actually belong to different genera their difference is just a matter of strength. For instance, one could think that one fully believes some proposition if one s confidence in it goes above a certain threshold, so that it makes sense to say that lower degrees of credence are just weaker forms of belief. It seems that situations analogous to MEDICINE can be easily found for these types of weak belief. The example of the coin above would be a case in point, but it is possible to devise examples with degrees of credence as high as desired. Imagine a fair lottery with 100 tickets. Before checking the result, I am permitted to have 0.99 credence that I have a losing ticket, even if I know that once I learn about the result this degree of credence will become impermissible (my new permissible degree of credence will be 1 if I find out that I have lost, 0 otherwise). One first thing to say is that it would still be the case that there are no MEDICINE-like cases for full-fledged, outright belief. If I take myself to being permitted to (fully) believe that p, then I cannot know that once I acquire all relevant evidence this believing attitude will be impermissible. Thus, the discussion in this paper would still reveal an interesting contrast between permissions to outright believing and permissions to act. A possible way of putting the contrast is that when the agent is not in a position to fulfil the demanding standards of outright believing, she may always resort to weaker epistemic attitudes that are not subject to such standards, whereas in general there are no weaker forms of acting available when one s evidence is limited (actions performed from such a limited perspective are no less full-fledged actions than those performed from more informed perspectives).

16 Furthermore, it seems to me that the sorts of considerations I have discussed actually offer reasons against thinking of outright belief as high credence. The lottery example shows that a permissible high degree of credence in p, if it is lower than 1, is compatible with acknowledging that such high credence may be impermissible from a further perspective with complete evidence without making one s current credence impermissible or mistaken. Arguably, even if p is actually false, my current high credence in it does not need to be mistaken, provided that p is very probable in light of my current evidence (your lucky lottery win does not imply that I was mistaken when regarding it as very probable that you would lose). Thus, if belief were high credence, it should be acceptable to say: I believe that p (since my credence in p is high enough), but the falsity of p would not necessarily make this believing attitude mistaken, because I am leaving room for p s being false. But this sounds wrong: an attitude that is not regarded as mistaken despite its content being false would not count as outright belief. High credence is tolerant to falsity in a way that belief is not. 15 So, my conclusion stands: in cases like MEDICINE (and plausibly in general) what is permissible to do is determined only by those reasons within one s epistemic reach, whereas what is permissible to (fully) believe is determined by all existing reasons. As a result of adding all relevant evidence to one s epistemic ken, something that was right to do (e.g. giving drug A) may become impermissible; in contrast, something that was permissible for the agent to (fully) believe (something true) would never become impermissible if she had access to all evidence (if she knew all facts). I have argued that the reason why this contrast takes place is that practical deliberation can only be brought to an end by endorsing some course of action whereas epistemic deliberation may be concluded by adopting attitudes other than full belief (perhaps some surrogate of belief with similar practical consequences, such as accepting a view as a working hypothesis, making a guess, or taking something as a sufficiently good approximation to the

17 right answer). There are situations in which not all relevant existing reasons are accessible to the agent, who nevertheless may be pressed to reach a conclusion in her deliberation. In the epistemic case, she may reach such a conclusion by adopting some attitude other than belief. In the practical case, in contrast, the agent s only choice is to endorse some course of action: in the context of practical deliberation, there are no surrogates for action. There is an urgency to deciding what to do that is not found in decisions about what to believe. Plausibly, this is why the normative standards governing the permissibility of actions depend on the agent s epistemic perspective: what one is permitted to do depends only on those reasons one has access to. What one is permitted to believe, in contrast, depends on all existing reasons on all facts. 5. Are Practical Perspectivism and Doxastic Objectivism incompatible? It may be thought that endorsing PP and DO at the same time is an unstable position. Does not the combination of PP and DO lead to incoherent views? In particular, it could seem that the combination of PP and DO makes it permissible for agents to have beliefs that are in tension with their practical permissions. Think again of MEDICINE. According to PP, Jill is permitted to prescribe drug A, even if the totality of facts supports prescribing drug B instead (if Jill had complete evidence, that would be the thing to do). But, does not DO, in turn, permit her to believe the claim that drug B is actually the cure and that therefore the totality of facts supports prescribing it? After all, such a claim is a truth. If Jill were permitted to believe that claim, it would be problematic, because according to DO she would be permitted to believe something that, once believed, would seem to offer permission to do something (giving drug B) which is impermissible according to PP. Note, however, that DO does not state that if an agent is permitted to believe

18 something from a perspective of complete evidence, she is also permitted to believe it from a perspective of limited evidence. That would amount to claiming that if a claim is true, then one is permitted to believe it. I am not committed to such a thesis. Rather, I have just assumed that one is permitted to believe something only if it is true. Thus, it does not follow that in MEDICINE Jill is entitled to believe that drug B is the cure, just because it is a true claim (even if she possesses no evidence supporting it). It is compatible with DO that one is permitted to believe something only if it is true and supported by one s evidence (say, only if known). A second possible instability is the following. I have said that an agent can lose her permission to perform some action A when she gets access to an enriched perspective of full evidence. For instance, Jill loses her permission to prescribe drug A when reaching a perspective with full evidence. But then, it would seem that Jill, from her initial impoverished perspective, could be permitted to believe that she is permitted to prescribe drug A (it is true and she knows it), whereas, from the enriched perspective, she would not be permitted to believe that she is permitted to prescribe drug A (it would be a false claim). Is not this a MEDICINE-like scenario for doxastic permissions? Jill seems to have lost permissions to believe by reaching a perspective of full evidence. This interpretation of Jill s situation is, however, mistaken. I have argued that what an agent is permitted to do depends on her epistemic perspective. Changes on the agent s perspective, therefore, will bring changes in her practical permissions. Practical permissions are relative to the epistemic perspective of the agent to whom such permissions are attributed. In this way, beliefs about Jill s practical permissions when she occupies an impoverished epistemic perspective are different from beliefs about her practical permissions when in a perspective of full evidence. So, when Jill, from a perspective of partial evidence believes that she is permitted to prescribe drug A, the content of her belief is compatible with what she believes when, from a perspective of full evidence, she believes that she is not permitted to

19 prescribe drug A. According to PP, those two beliefs are about different permissions (since they are about situations in which Jill occupies different epistemic perspectives, and PP holds that practical permissions are relative to such perspectives). Of course, PP does not deny that Jill, looking back from her perspective of full evidence, is permitted to believe that she was permitted to prescribe drug A (when she occupied a perspective of partial evidence). This belief is perfectly compatible, according to PP, with the belief that Jill is now (occupying her perspective of full evidence) not permitted to prescribe drug A. So, there does not seem to be any obvious instability in the combination of PP and DO. But, one may wonder, is it not preferable to have a unified account in which both practical and doxastic permissions are either perspectivist or objectivist? I do not think that this is so, if one has a good explanation of why the two types of permission behave differently an explanation that I have tried to provide in this paper. Perhaps there is some prima facie appeal in unified accounts, but surely this appeal is reduced when there is a convincing story about why the phenomena studied do not manifest such unity. Sometimes unified views just do not fit the facts. 6. Conclusions I have argued that there is a relevant dissimilarity between the normative standards governing beliefs and actions. Permissions to act possessed in a perspective with limited evidence may be lost when reaching a perspective with complete evidence, whereas this cannot happen with permissions to believe. Such a dissimilarity becomes particularly clear in scenarios like MEDICINE, in which it is permissible for an agent to do something that she knows would be impermissible for her if she knew all facts. It is impossible to find this sort of cases for belief: there are no situations in which it is permissible for one to believe what one knows to be false.

20 I have proposed that the source of this dissimilarity is that practical deliberations can only be closed by endorsing some course of action, while epistemic deliberations can be concluded by adopting attitudes other than belief including surrogates of belief with similar practical consequences, such as accepting some view as a good approximation. Thus, even when there are pressures to bring some epistemic deliberation to an end, the agent will always be in a position to refrain from forming a belief if she knows there is not sufficient evidence for doing so: she can always conclude her deliberation by adopting some other suitable attitude. In contrast, if an agent is pressed to conclude some practical deliberation, she can only do so by endorsing a course of action. This explains why practical permissions are sensitive only to reasons within the agent s epistemic ken. Acknowledgements I am grateful to Conor McHugh, Jesús Vega, Daniel Whiting and two anonymous referees for their helpful comments. I also want to thank the audience at the VIII meeting of the Spanish Society of Logic, Methodology and Philosophy of Science, in Barcelona, where a previous version of this paper was presented. This work was partly supported by the Spanish Government research grant FFI P.

21 Notes 1. One may think that believing is something we do (perhaps even some type of mental action). If so, this paper can be read as an exploration of a contrast between permissions regarding two different kinds of doings: on the one hand, permissions to form beliefs, on the other hand permissions to perform non-doxastic actions. In what follows, when I talk of permissions to do something I will be referring to non-doxastic doings (i.e. to doing things other than forming a belief). 2. Plausibly, some attitude or action is sufficiently supported by the balance of reasons if the reasons in its favour are at least as weighty as the reasons against it, or in favour of alternatives (see Schroeder 2015a). 3. Perspectivist positions have been endorsed by authors such as Lord (2015), Kiesewetter (2011), Parfit (2011), Zimmerman (2006), Kolodny and MacFarlane (2010) and Dowell (2013). 4. Objectivist views are defended among others by Graham (2010), Bykvist (2011) and Thomson (1986). 5. My answer to this question is negative. I think that only real, actually good reasons determine practical obligations and permissions. 6. What determines what evidence is considered available to the agent what facts the agent is supposed to acquire knowledge about? If the answer is that the reasons within her epistemic ken, it seems we will enter a regress (for I was trying to explain precisely what being within the agent s ken amounts to). There are several ways of tackling this issue (see Kiesewetter 2011 for discussion). For the purposes of this essay, it may be assumed that what information the agent is supposed to acquire is determined by the agent s possessed reasons more specifically, by the agent s possessed evidence about her epistemic environment (i.e. about where relevant information may be found) and about the importance of her decision. For instance, if I have reasons to think that relevant information is likely to be found in a book in

22 the library, I should, prima facie, try to read the book. Moreover, the importance of the decision will determine how hard the agent has to look for evidence (e.g. if I possess reasons to suspect that the life of people depends on my decision, I will have reasons to look harder for information than if my decision had lesser consequences). In general, the agent ought to try to acquire as much evidence as necessary for making an informed decision, within the limits imposed by her abilities, by time constraints and by the balance between the importance of the decision and the costs of the search for evidence. 7. Although this is not uncontroversial, see Thomson (2008) and Glüer and Wikforss (2009). It is worth noting that I do not need to commit myself to the view that correctness can be reduced to permissibility. For my purposes here, it is enough to assume that incorrect beliefs are impermissible. It also important to point out that, while on the views I will explore holding a false belief is impermissible, it can be a perfectly rational thing to do if it was supported by what appeared to be good reasons (for the distinction between doing something impermissible or incorrect and doing something irrational, see Williamson forthcoming; Schroeder 2015b; Whiting 2014; Sylvan forthcoming). 8. I will not consider here the complementary view, that is, the claim that if something is true it is permissible for one to believe it (see Whiting 2010, 2013). 9. An interesting question is how these considerations are affected by the possibility of relativist beliefs, in particular beliefs relativized to epistemic perspectives (i.e. beliefs whose truth depends not only on the world of evaluation but also on epistemic perspectives). A possible candidate is epistemic modals (see MacFarlane 2014). Take (1): (1) James may be in London, but he may also be in Bristol Arguably, (1) is true relative to a body of evidence e1 that leaves open the possibility that James is in London and also that he is in Bristol, while it is false relative to an enriched body of evidence e2 ruling out at least one of those possibilities. Thus, an agent may be permitted to believe (1) from perspective e1, even if she knows that from a perspective of complete evidence it is false.

23 Of course, the relativist picture of epistemic modals can be disputed. Perhaps better accounts are offered by contextualism (Kratzer 2012; Dowell 2011) according to which (1) would express different beliefs from different perspectives, or expressivism (Yalcin 2011; Schnieder 2010) according to which speech involving epistemic modals does not express beliefs, but some other attitude. I focus on non-relativist beliefs for the remaining of the paper. 10. For a discussion of reasons for withholding belief, see Schroeder (2012). 11. I will consider that forming an intention to act, or even making a normative judgement recommending the performance of the action, count as ways of endorsing a course of action. So, I remain neutral on the issue whether practical deliberation ends in an action, the formation of an intention or the endorsement of a normative judgment. 12. Nothing of what I am saying hangs, either, on whether suspending judgement counts as a way of concluding epistemic deliberations or only as a way of deferring such conclusion (see Friedman 2013). 13. Perhaps there are irresoluble practical dilemmas in which there is no rational way for the agent to choose among alternative actions (even if she is forced to make a decision). In these situations, it seems, the agent would be without rational guidance. But this is not what happens in scenarios like MEDICINE: intuitively, the agent can claim to have good reasons to give the safe drug rather than doing nothing or taking reckless risks. 14. In particular, it would be wrong to assign 50% probability to the coin having landed heads up if one knows that the coin actually did land tails up. 15. Ross and Schroeder (2014) discuss arguments along these lines against the view of belief as high credence. Notes on the contributor Javier González de Prado Salas received his PhD from the University of Southampton in January 2016 and is now a Teaching Fellow at Universidad Autónoma de Madrid (Spain). He works on topics in normativity theory, epistemology, philosophy of language and aesthetics. He is particularly interested

24 in investigating the nature of defeasible reasoning and its implications for different philosophical debates. References Alvarez, María Kinds of reasons: an essay in the philosophy of action. Oxford: Oxford University Press. Bykvist, Krister How to Do Wrong Knowingly and Get Away with It. In Neither/Nor. Philosophical Papers Dedicated to Erik Carlson on the Occasion of His Fiftieth Birthday, edited by Rysiek Sliwinski and Frans Svensson, Uppsala: Department of Philosophy, Uppsala University. Boghossian, Paul Content and Justification. Oxford: Oxford University Press. Dorsey, Dale Objective Morality, Subjective Morality and the Explanatory Question. Journal of Ethics & Social Philosophy 6: Dowell, Janice A Flexible Contextualism Account of Epistemic Modals. Philosophers Imprint 11: Dowell, Janice Flexible Contextualism about Deontic Modals: A Puzzle about Information-Sensitivity. Inquiry 56: Feldman, Richard Epistemic obligations. Philosophical Perspectives 2: Friedman, Jane Suspended judgment. Philosophical studies 162: Glüer, Kathrin and Asa Wikforss Against content normativity. Mind, 118: Graham, Peter A In defense of objectivism about moral obligation. Ethics 121:

25 Jackson, Frank Decision-theoretic consequentialism and the nearest-dearest objection. Ethics 101: Kiesewetter, Benjamin Ought and the perspective of the agent. Journal of Ethics and Social Philosophy 5: Kolodny, Niko and John MacFarlane Ifs and oughts. The Journal of philosophy 107: Kratzer, Angelika Modals and Conditionals: New and Revised Perspectives. Oxford: Oxford University Press. Littlejohn, Clayton Justification and the Truth-Connection. Cambridge: Cambridge University Press. Lord, Errol Acting for the right reasons, abilities, and obligation. In Oxford Studies in Metaethics, vol. 10, edited by Russ Shafer-Landau, Oxford University Press. MacFarlane, John Assessment sensitivity: Relative truth and its applications. Oxford: Oxford University Press. Parfit, Derek On What Matters. Oxford: Oxford University Press. Ross, Jacob and Mark Schroeder Belief, credence, and pragmatic encroachment. Philosophy and Phenomenological Research 88: Scanlon, Thomas M What we owe to each other. Cambridge, MA: Harvard University Press. Schnieder, Benjamin Expressivism concerning epistemic modals. The Philosophical Quarterly 60: Schroeder, Mark Stakes, withholding, and pragmatic encroachment on knowledge. Philosophical studies 160:

26 Schroeder, Mark. 2015a. What Makes Reasons Sufficient? American Philosophical Quaterly 52: Schroeder, Mark. 2015b. Knowledge is Belief for Sufficient (Objective and Subjective) Reason. In Oxford Studies in Epistemology: Volume 5, edited by Tamar Szabó Gendler and John Hawthorne, Oxford: Oxford University Press. Sepielli, Andrew. Forthcoming. Subjective and Objective Reasons. In Oxford Handbook on Reasons and Normativity, edited by Daniel Star. Oxford: Oxford University Press. Shah, Nishi and J. David Velleman Doxastic Deliberation. Philosophical Review 114: Sylvan, Kurt Epistemic Reasons I: Normativity. Philosophy Compass 11: Thomson, Judith J Imposing Risks. In Rights, restitution, and risk: Essays in moral theory, edited by William Parent, Cambridge, MA: Harvard University Press. Thomson, Judith J Normativity. Chicago: Open Court. Whiting, Daniel Should I Believe the Truth? Dialectica 64: Whiting, Daniel Keep Things in Perspective: Reasons, Rationality and the A Priori. Journal of Ethics & Social Philosophy 8: Williamson, Timothy Knowledge and its Limits. Oxford: Oxford University Press. Williamson, Timothy. Forthcoming. Justifications, excuses, and sceptical scenarios. In The New Evil Demon, edited by Fabian Dorsch and Julien Dutant. Oxford: Oxford University Press. Yalcin, Seth Nonfactualism about epistemic modality. In Epistemic modality, edited by Aandy Egan and Brian Weatherson, Oxford: Oxford University Press.